(585d) A High-Throughput in Vitrocompartmentalization (IVC) Directed Evolution Platform for Engineering Protease Substrate Specificity

Genes encoding proteases comprises ~2 of the human genome. Through their ability to catalytically activate and inactivate their protein targets, proteases regulate the fate, localization, and activity of many proteins, create new bioactive molecules, processes of cellular information, and generate, transduce, and amplify molecular signals. Due to this catalytic nature, proteases present unique opportunities for use in therapeutic applications, in ways that could complement, replace existing therapies or even introduce treatments for diseases with no current cure. The idea of using engineered proteases for therapeutic applications has long been a goal in the biopharma industry. There are currently twelve FDA-approved recombinant protease drugs on the market for several blood-related disorders and muscle spasms, with several more proteases currently in various phases of clinical trial. Unfortunately, all current recombinant protease therapies have been restricted to a handful of proteases with narrow and native substrate specificities, since more promiscuous proteases are highly toxic. As a result, taking advantage of the untapped potential of proteases as therapeutics is contingent on the development of methods that engineer protease specificity in ways that narrow the specificity of promiscuous proteases or that completely repurpose their specificity towards completely new targets. For instance, proteases with fine-tuned specificity could be envisioned as therapeutics in diseases such as chronic inflammation, neurodegenerative, autoimmune diseases and cancer, leveraging their unique ability to degrade disease-associated proteins. For instance, proteases hold promise as a treatment for IgG-mediated pathogenic conditions (rheumatoid arthritis, systemic lupus erythematosus, multiple myeloma). Recent evidence showed that ablating IgG levels through cleavage of IgG hinges by the bacterial enzyme IdeS (immunoglobulins degrading enzyme from Streptococcus pyogenes) alleviates autoimmune disease symptoms in a number of animal models of antibody-mediated disorders and reduces transplant rejection in humans. Although effective, IdeS inevitably triggers an immune reaction that diminishes or abrogates its efficacy. Therefore, we hypothesize that an engineered human protease would be preferred for repeated injections and possibly lead to an effective long-term therapeutic.

In this work, we present a novel high-throughput IVC directed evolution platform for engineering proteases. In this system, a genotype-phenotype linkage between a DNA encoding a protease of interest and an activity reporter (a protein or peptide substrate) is established on the surface of a microbead. The reporter contains epitope tags attached to the N- and C- termini of the protein substrate or flanking protease cleavage sites within a peptide substrate. Up to 10⁷ microbead-DNA-protein substrate complexes are individually encapsulated in water-in-oil picoliter droplets along with an in vitrotranscription-translation system, wherein the expressed protease cleaves the substrate reporter. De-emulsified microbeads are stained with fluorescently-labeled epitope-specific antibodies and sorted by fluorescence-activated cell sorting (FACS), thereby isolating beads bearing protease variants with higher activity or switched specificity. Compared with other high-throughput protease engineering methods, our in vitroplatform offers the advantages of 1) bypassing protease-induced cytotoxicity, 2) enabling engineering on full proteins in the context of their 3D structures rather than peptide substrates and 3) offering a level of control over screening and expression conditions not achievable in vivo. Furthermore, thisin vitro system offers the capacity to screen large DNA libraries (up to 10⁷) in less than one day. After optimization, we tested our system with the immunoglobulin degrading enzyme IdeS and an IgG1 reporter. We show that active IdeS mixed with inactive IdeS at a 1:10000 ratio can be fully recovered after two rounds of FACS sorting. This is particularly noteworthy because IdeS is only active on full IgG and has shown no activity on peptide substrates. This makes our system the first capable of engineering IdeS and related Ig-degrading proteases. When applied to a library of MMP3, we have isolated MMP3 mutants that show higher activity on IgG1-hinge cleaving. Lastly, we have used this platform to screen activities of other matrix metalloproteases (MMPs 1, 3, 7, 8, 12, 13). Using a golden gate cloning approach to create a library of MMP chimeras, we can quickly interrogate the contribution of substrate determining residues found in flexible loops of each MMP to overall MMP substrate specificity.